flexablelib 2.0.18__tar.gz → 2.0.20__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (59) hide show
  1. {flexablelib-2.0.18 → flexablelib-2.0.20}/CHANGELOG.md +60 -0
  2. {flexablelib-2.0.18/flexablelib.egg-info → flexablelib-2.0.20}/PKG-INFO +2 -2
  3. {flexablelib-2.0.18 → flexablelib-2.0.20}/README.md +1 -1
  4. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/cells.py +19 -4
  5. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/colors.py +42 -10
  6. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/columns.py +11 -2
  7. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/merges.py +21 -5
  8. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/readers.py +140 -15
  9. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/table.py +167 -50
  10. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/table_strings.py +42 -2
  11. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/writers.py +8 -0
  12. {flexablelib-2.0.18 → flexablelib-2.0.20/flexablelib.egg-info}/PKG-INFO +2 -2
  13. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib.egg-info/SOURCES.txt +3 -0
  14. {flexablelib-2.0.18 → flexablelib-2.0.20}/package.json +1 -1
  15. flexablelib-2.0.20/tests/test_v2110_fixes.py +133 -0
  16. flexablelib-2.0.20/tests/test_v219_coverage.py +162 -0
  17. flexablelib-2.0.20/tests/test_v219_fixes.py +480 -0
  18. {flexablelib-2.0.18 → flexablelib-2.0.20}/LICENSE +0 -0
  19. {flexablelib-2.0.18 → flexablelib-2.0.20}/MANIFEST.in +0 -0
  20. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/__init__.py +0 -0
  21. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/_speedups.c +0 -0
  22. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/_width_tables.h +0 -0
  23. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/fast.py +0 -0
  24. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/options.py +0 -0
  25. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/py.typed +0 -0
  26. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/style_compositions.py +0 -0
  27. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/text_width.py +0 -0
  28. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib/utils.py +0 -0
  29. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib.egg-info/dependency_links.txt +0 -0
  30. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib.egg-info/requires.txt +0 -0
  31. {flexablelib-2.0.18 → flexablelib-2.0.20}/flexablelib.egg-info/top_level.txt +0 -0
  32. {flexablelib-2.0.18 → flexablelib-2.0.20}/pytest.ini +0 -0
  33. {flexablelib-2.0.18 → flexablelib-2.0.20}/setup.cfg +0 -0
  34. {flexablelib-2.0.18 → flexablelib-2.0.20}/setup.py +0 -0
  35. {flexablelib-2.0.18 → flexablelib-2.0.20}/style_examples.md +0 -0
  36. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/conftest.py +0 -0
  37. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/helpers.py +0 -0
  38. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_cells.py +0 -0
  39. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_columns.py +0 -0
  40. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_excel.py +0 -0
  41. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_features.py +0 -0
  42. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_issues.py +0 -0
  43. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_merge_edge_matrix.py +0 -0
  44. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_merges.py +0 -0
  45. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_readme_examples.py +0 -0
  46. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_styles.py +0 -0
  47. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_text_width.py +0 -0
  48. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_two_pass_measurement.py +0 -0
  49. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_upstream_issues.py +0 -0
  50. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v213_features.py +0 -0
  51. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v214_features.py +0 -0
  52. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v215_features.py +0 -0
  53. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v216_features.py +0 -0
  54. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v217_features.py +0 -0
  55. {flexablelib-2.0.18 → flexablelib-2.0.20}/tests/test_v218_features.py +0 -0
  56. {flexablelib-2.0.18 → flexablelib-2.0.20}/tools/benchmark.py +0 -0
  57. {flexablelib-2.0.18 → flexablelib-2.0.20}/tools/generate_width_tables.py +0 -0
  58. {flexablelib-2.0.18 → flexablelib-2.0.20}/tools/readme_gallery.py +0 -0
  59. {flexablelib-2.0.18 → flexablelib-2.0.20}/tools/text_to_png.py +0 -0
@@ -2,6 +2,66 @@
2
2
 
3
3
  All notable changes to this project will be documented in this file. See [standard-version](https://github.com/conventional-changelog/standard-version) for commit guidelines.
4
4
 
5
+ ## [2.0.20] - 2026-11-02
6
+
7
+ ### Fixed
8
+ - ``column_min_width`` and ``column_max_width`` are checked when set. A
9
+ non-integer entry raised a bare ``int()`` error only at print time, and a
10
+ cap of ``0`` or below silently trimmed every cell to ``.``. Both now raise
11
+ ``ValueError`` naming the option; ``None`` still means no bound.
12
+ - ``from_html`` with a negative ``index`` past the first table raised a bare
13
+ ``list index out of range``; it now gives the same message as a positive
14
+ index that is out of range.
15
+
16
+ ### Tests
17
+ - Reader coverage: CSV without header and ragged rows, dialect plus
18
+ delimiter, numeric-looking text, Markdown from bytes and paths, HTML from
19
+ paths, records with dict rows and generator input.
20
+
21
+ ## [2.0.19] - 2026-11-01
22
+
23
+ ### Fixed
24
+ - ``column_max_width`` trims by terminal columns and line by line: CJK text
25
+ is cut to fit, emoji ZWJ sequences are no longer over-trimmed and
26
+ multi-line cells keep their lines instead of collapsing to ``...``.
27
+ - A width cap never narrows a column below content that cannot shrink (the
28
+ index column, a float's integer digits, a 2-column glyph in a 1-column
29
+ cap); the frame used to come out narrower than the text.
30
+ - ``show_index`` with ``column_max_width`` raised ``IndexError`` or trimmed
31
+ the wrong column.
32
+ - Float columns fill wider columns (``column_min_width``, a long header)
33
+ instead of leaving rows short.
34
+ - A numeric-looking header (``'0'``, ``'2024'``) over a float column crashed
35
+ with ``valign`` and overflowed without it; it is now right-aligned.
36
+ - ``sort_by`` on mixed types or ``None`` raised ``TypeError``. Numbers sort
37
+ before text, other types group by type, and missing values (``None``,
38
+ NaN) always go last, also with ``sort_reverse``.
39
+ - ``border_color`` and ``border_text`` emitted ANSI escapes with
40
+ ``use_colors=False``, under ``NO_COLOR`` or into a pipe (regression from
41
+ 2.0.11). Coloured themes now leave ``use_colors`` on auto instead of
42
+ forcing it on, so ``apply_theme`` output to a file stays clean; set
43
+ ``use_colors = True`` to force colour.
44
+ - Left/right ``border_text`` overwrote cell text on styles without an outer
45
+ border, and started on the title line when a ``title`` was set.
46
+ - ``sticks`` alignment indicators overhung columns narrower than 3.
47
+ - ``add_divider`` on styles without a body rule (``dashes``,
48
+ ``grid_eheader``) drew a rule too wide after a wrapped row.
49
+ - Merged cells follow ``cell_padding`` (0, 2, 3...); ``hrules`` with merges
50
+ on styles whose frame has no horizontal glyph raised ``TypeError``.
51
+ - Three or more repeated headers (``['x', 'x', 'x']``) produced a duplicate
52
+ name and ``compose()`` raised ``IndexError``.
53
+ - Readers: ``from_markdown`` unescapes ``\|``; ``from_records`` accepts the
54
+ dicts ``to_records`` returns; ``from_html`` reads ``to_html`` pages, lays
55
+ out ``rowspan``/``colspan`` (later cells used to slide left) and restores
56
+ them as merges, and takes the ``<th>`` header from the selected table.
57
+ - HTML/XLSX colours: a background-only ``'on red'`` exported as red text,
58
+ and ``greyNN`` / ``color:N`` colours exported nothing.
59
+ - ``to_latex`` escapes ``|``, ``<`` and ``>``.
60
+
61
+ ### Tests
62
+ - Regression tests for every fix above and behaviour tests for exporters,
63
+ readers and colour helpers (coverage 87% -> 90%).
64
+
5
65
  ## [2.0.18] - 2026-10-31
6
66
 
7
67
  ### Fixed
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: flexablelib
3
- Version: 2.0.18
3
+ Version: 2.0.20
4
4
  Summary: Fast, dependency-free terminal tables for Python: 56 border styles, ANSI colors, cell merges, decimal alignment, width fitting, and CSV/Markdown/HTML/Excel/LaTeX I/O.
5
5
  Home-page: https://github.com/Opus-Perpetuus/flexablelib
6
6
  Author: Benjamin Ramirez
@@ -204,7 +204,7 @@ one place.
204
204
  tabulate and ~18× faster than rich's table. pandas `to_string()` is faster
205
205
  still (~85 ms). Full method and numbers:
206
206
  [docs/BENCHMARKS.md](https://github.com/Opus-Perpetuus/flexablelib/blob/main/docs/BENCHMARKS.md).
207
- - **Wheels.** Prebuilt for CPython 3.8 – 3.13 on Linux (manylinux, musllinux),
207
+ - **Wheels.** Prebuilt for CPython 3.8 – 3.14 on Linux (manylinux, musllinux),
208
208
  Windows x64 and macOS arm64. Anywhere else it builds from source or falls
209
209
  back to pure Python — installation never fails for want of a compiler.
210
210
  `from flexablelib.fast import implementation; implementation()` tells you
@@ -139,7 +139,7 @@ one place.
139
139
  tabulate and ~18× faster than rich's table. pandas `to_string()` is faster
140
140
  still (~85 ms). Full method and numbers:
141
141
  [docs/BENCHMARKS.md](https://github.com/Opus-Perpetuus/flexablelib/blob/main/docs/BENCHMARKS.md).
142
- - **Wheels.** Prebuilt for CPython 3.8 – 3.13 on Linux (manylinux, musllinux),
142
+ - **Wheels.** Prebuilt for CPython 3.8 – 3.14 on Linux (manylinux, musllinux),
143
143
  Windows x64 and macOS arm64. Anywhere else it builds from source or falls
144
144
  back to pure Python — installation never fails for want of a compiler.
145
145
  `from flexablelib.fast import implementation; implementation()` tells you
@@ -186,6 +186,21 @@ def fljust(string: str,
186
186
  return ''.join([left, point or fill_char, right])
187
187
 
188
188
 
189
+ def __fill_to(aligned: str, cell_length: int, fill_char: str) -> str:
190
+ """
191
+ Right-align a decimal-aligned block within a wider column.
192
+
193
+ The block is exactly as wide as the column's widest number. When the
194
+ column is wider than that -- a ``column_min_width``, a long header, or
195
+ a text cell such as ``None`` -- the numbers came out short and the row
196
+ border moved left. Padding every block on the left by the same amount
197
+ keeps the decimal points on one axis.
198
+ """
199
+ if visible_width(aligned) < cell_length:
200
+ return pad_to_width(aligned, cell_length, 'r', fill_char)
201
+ return aligned
202
+
203
+
189
204
  def __fljust_part(cell_part: Any,
190
205
  cell_length: int,
191
206
  float_column_sizes: Union[List[int], Tuple[int, int, int]],
@@ -228,11 +243,11 @@ def __fljust_part(cell_part: Any,
228
243
 
229
244
  # If it is a float, align it as a float.
230
245
  if is_float_number:
231
- return fljust(
246
+ return __fill_to(fljust(
232
247
  str_cell,
233
248
  float_column_sizes,
234
249
  fill_char
235
- )
250
+ ), cell_length, fill_char)
236
251
 
237
252
  # If it is an int, align it as an int.
238
253
  # Will be aligned to the int part of the float.
@@ -266,13 +281,13 @@ def __fljust_part(cell_part: Any,
266
281
  fixed_right_width = right_width + float_column_sizes[POINT_WIDTH_I]
267
282
 
268
283
  # Add the spaces to the left and right.
269
- return _add_cell_spacing(
284
+ return __fill_to(_add_cell_spacing(
270
285
  str_cell,
271
286
  fixed_left_width,
272
287
  fixed_right_width,
273
288
  0,
274
289
  fill_char
275
- )
290
+ ), cell_length, fill_char)
276
291
 
277
292
  # If it is not a float nor an int, align it to the right.
278
293
  else:
@@ -269,7 +269,12 @@ def _split_spec(spec: Union[str, int, None]):
269
269
  return [], None
270
270
  if isinstance(spec, int):
271
271
  return [], None
272
- foreground_part, _, background_part = str(spec).lower().partition(' on ')
272
+ lowered = str(spec).lower().strip()
273
+ # Same grammar as parse_spec: a bare 'on <bg>' is background-only.
274
+ if lowered.startswith('on '):
275
+ foreground_part, background_part = '', lowered[3:]
276
+ else:
277
+ foreground_part, _, background_part = lowered.partition(' on ')
273
278
  fg_tokens = foreground_part.split()
274
279
  bg_token = None
275
280
  if background_part:
@@ -278,15 +283,44 @@ def _split_spec(spec: Union[str, int, None]):
278
283
  return fg_tokens, bg_token
279
284
 
280
285
 
286
+ def _xterm_hex(index: int) -> Optional[str]:
287
+ """sRGB hex for an xterm 256-colour index (16..255; 0..15 are themed)."""
288
+ if 16 <= index <= 231:
289
+ index -= 16
290
+ steps = [0, 95, 135, 175, 215, 255]
291
+ rgb = (steps[index // 36], steps[(index // 6) % 6], steps[index % 6])
292
+ elif 232 <= index <= 255:
293
+ level = 8 + 10 * (index - 232)
294
+ rgb = (level, level, level)
295
+ else:
296
+ return None
297
+ return '{0:02X}{1:02X}{2:02X}'.format(*rgb)
298
+
299
+
300
+ def _token_hex(token: str, named: dict) -> Optional[str]:
301
+ """Hex for one colour token: #rrggbb, a name, greyNN or a 256 index."""
302
+ hex_match = _HEX_RE.match(token)
303
+ if hex_match:
304
+ return hex_match.group(1).upper()
305
+ if token in named:
306
+ return named[token]
307
+ grey_match = _GREY_RE.match(token)
308
+ if grey_match and int(grey_match.group(1)) <= 100:
309
+ return _xterm_hex(_grey_index(int(grey_match.group(1))))
310
+ index_match = _INDEX_RE.match(token)
311
+ if index_match:
312
+ return _xterm_hex(int(index_match.group(1)))
313
+ if token.isdigit():
314
+ return _xterm_hex(int(token))
315
+ return None
316
+
317
+
281
318
  def background_hex(spec: Union[str, int, None]) -> Optional[str]:
282
319
  """sRGB hex (no ``#``) for the background half of a colour spec, if any."""
283
320
  _fg, bg = _split_spec(spec)
284
321
  if bg is None:
285
322
  return None
286
- hex_match = _HEX_RE.match(bg)
287
- if hex_match:
288
- return hex_match.group(1).upper()
289
- return _BG_HEX.get(bg)
323
+ return _token_hex(bg, _BG_HEX)
290
324
 
291
325
 
292
326
  def foreground_hex(spec: Union[str, int, None]) -> Optional[str]:
@@ -295,11 +329,9 @@ def foreground_hex(spec: Union[str, int, None]) -> Optional[str]:
295
329
  for token in fg_tokens:
296
330
  if token in ATTRIBUTES:
297
331
  continue
298
- hex_match = _HEX_RE.match(token)
299
- if hex_match:
300
- return hex_match.group(1).upper()
301
- if token in _FG_HEX:
302
- return _FG_HEX[token]
332
+ found = _token_hex(token, _FG_HEX)
333
+ if found:
334
+ return found
303
335
  return None
304
336
 
305
337
 
@@ -852,8 +852,12 @@ def __align_one_header(column: Union[str, list, tuple],
852
852
  column
853
853
  ))
854
854
  else:
855
- # Apply to the cell.
856
- yield __align_single_cell(column, col_width, to_where, margin)
855
+ # Apply to the cell. The float sizes must be passed here too: a
856
+ # numeric-looking header ('0', '2024') over a float column was
857
+ # decimal-aligned with no sizes and raised TypeError.
858
+ yield __align_single_cell(
859
+ column, col_width, to_where, margin, float_column_sizes
860
+ )
857
861
 
858
862
 
859
863
  def _align_headers(style_composition: TableComposition,
@@ -875,6 +879,11 @@ def _align_headers(style_composition: TableComposition,
875
879
  for column_i, column in enumerate(headers):
876
880
  # Get the alignment of the current column.
877
881
  to_where = col_alignments[column_i]
882
+ if to_where == COLUMN_ALIGNS.float:
883
+ # A header is a label, not a number. Decimal-aligning one that
884
+ # looks numeric ('2024') pushed it past the column edge; text
885
+ # headers already came out right-aligned, so all of them do.
886
+ to_where = COLUMN_ALIGNS.right
878
887
 
879
888
  # Get the width that the current column should have.
880
889
  col_width = col_widths[column_i]
@@ -166,7 +166,7 @@ def compute_layout(column_widths, composition, hidden=()) -> Layout:
166
166
  so counting them here would put every offset after them too far right.
167
167
  """
168
168
  body = composition.vertical_table_body_lines
169
- margin = 1 if composition.margin else 0
169
+ margin = max(0, int(composition.margin or 0))
170
170
  hidden = set(hidden)
171
171
 
172
172
  offset = len(body.left) if body.left is not None else 0
@@ -227,6 +227,10 @@ def _region_bounds(region, layout, index_offset):
227
227
  _H_CHARS = set('─━═┄┅┈┉╌╍▬−' '-~=')
228
228
  # A real rule has a *run* of dashes. A lone '-' inside "Beta-1" is not a rule.
229
229
  _H_RUN = re.compile(r'[─━═┄┅┈┉╌╍▬−\-~=]{2,}')
230
+ # Corner and tee glyphs a rule may carry; plain verticals are left out.
231
+ _JUNCTIONS = set('+') | {
232
+ chr(c) for c in range(0x2500, 0x2580)
233
+ } - set('│┃║┊┋┆┇╎╏')
230
234
  _PURE_H = {(0, 0, 1, 1), (0, 0, 1, 0), (0, 0, 0, 1)}
231
235
  _ROUND_CORNERS = set('╭╮╰╯')
232
236
 
@@ -362,7 +366,7 @@ def _synthesize_body_rule(composition):
362
366
  if v.middle in '|+':
363
367
  return SeparatorLine(
364
368
  left='+' if v.left else None,
365
- middle=h if h in '-=~' else '-',
369
+ middle=h if h and h in '-=~' else '-',
366
370
  intersection='+',
367
371
  right='+' if v.right else None,
368
372
  )
@@ -394,7 +398,7 @@ def _build_rule_line(layout, composition, column_widths, hidden, rule=None):
394
398
  if rule is None:
395
399
  return None
396
400
  v = composition.vertical_table_body_lines
397
- margin = 1 if composition.margin else 0
401
+ margin = max(0, int(composition.margin or 0))
398
402
  hidden = set(hidden or ())
399
403
  chunks = []
400
404
  first = True
@@ -470,7 +474,19 @@ def _looks_like_rule(line: str, h_glyph: str = None) -> bool:
470
474
  return False
471
475
  if h_glyph and len(h_glyph) >= 1 and (h_glyph * 2) in line:
472
476
  return True
473
- return _H_RUN.search(line) is not None
477
+ if _H_RUN.search(line) is not None:
478
+ return True
479
+ # cell_padding=0 with 1-column cells draws rules like '├─┼─┤', with no
480
+ # run of two dashes (and a merge may have blanked part of it). Such a
481
+ # line is a rule when it holds dashes and junctions and nothing else but
482
+ # spaces; a body line always has text or a plain vertical.
483
+ plain = strip_ansi(line)
484
+ return (
485
+ any(ch in _H_CHARS for ch in plain)
486
+ and any(ch in _JUNCTIONS for ch in plain)
487
+ and all(ch in _H_CHARS or ch in _JUNCTIONS or ch == ' '
488
+ for ch in plain)
489
+ )
474
490
 
475
491
 
476
492
  def _repair_rule_junctions(line, layout, composition, h_glyph, missing_south=None, missing_north=None,
@@ -574,7 +590,7 @@ def apply(lines, row_line_map, regions, column_widths, composition,
574
590
 
575
591
  layout = compute_layout(column_widths, composition, hidden_columns)
576
592
  lines = list(lines)
577
- margin = 1 if composition.margin else 0
593
+ margin = max(0, int(composition.margin or 0))
578
594
  body_rule = composition.table_body_line
579
595
  h_glyph = body_rule.middle if body_rule is not None else None
580
596
  # Separator lines whose horizontal run was blanked by a vertical span.
@@ -28,6 +28,7 @@ import csv as _csv
28
28
  import io as _io
29
29
  import re
30
30
  from html.parser import HTMLParser
31
+ from collections.abc import Mapping
31
32
  from typing import Any, Dict, Iterable, List, Optional, Sequence
32
33
 
33
34
  # Numeric text that should become a number. Deliberately strict: no leading
@@ -101,6 +102,16 @@ def from_records(table_class, rows: Iterable[Sequence],
101
102
 
102
103
  With no headers, the first row is used as the header row.
103
104
  """
105
+ rows = list(rows)
106
+ if rows and all(isinstance(row, Mapping) for row in rows):
107
+ # ``list(dict)`` is its keys, so dict rows (what ``to_records``
108
+ # returns) used to become a table of repeated header names.
109
+ if headers is None:
110
+ return from_dicts(table_class, rows, parse_numbers=parse_numbers,
111
+ **options)
112
+ headers = list(headers)
113
+ rows = [[row.get(key, '') for key in headers] for row in rows]
114
+ return _new_table(table_class, headers, rows, parse_numbers, **options)
104
115
  rows = [list(row) for row in rows]
105
116
  if headers is None:
106
117
  if not rows:
@@ -220,6 +231,11 @@ class _TableExtractor(HTMLParser):
220
231
  self._current_cell = None
221
232
  self._header_flags: List[bool] = []
222
233
  self._row_is_header = False
234
+ # Per table: which rows are <th> rows, and the rowspan/colspan blocks
235
+ # as (row, column, last_row, last_column) in the expanded grid.
236
+ self.header_flags: List[List[bool]] = []
237
+ self.spans: List[List[tuple]] = []
238
+ self._cell_span = (1, 1)
223
239
 
224
240
  def handle_starttag(self, tag, attrs):
225
241
  if tag == 'table':
@@ -235,13 +251,17 @@ class _TableExtractor(HTMLParser):
235
251
  self._row_is_header = False
236
252
  elif tag in ('td', 'th'):
237
253
  self._current_cell = []
254
+ self._cell_span = (_span(attrs, 'rowspan'), _span(attrs, 'colspan'))
238
255
  if tag == 'th':
239
256
  self._row_is_header = True
240
257
 
241
258
  def handle_endtag(self, tag):
242
259
  if tag == 'table':
243
260
  if self._table_depth == 1 and self._current_table is not None:
244
- self.tables.append(self._current_table)
261
+ grid, spans = _expand_spans(self._current_table)
262
+ self.tables.append(grid)
263
+ self.spans.append(spans)
264
+ self.header_flags.append(list(self._header_flags))
245
265
  self._current_table = None
246
266
  self._table_depth = max(0, self._table_depth - 1)
247
267
  return
@@ -252,7 +272,7 @@ class _TableExtractor(HTMLParser):
252
272
  # Collapse whitespace the way a browser would render it.
253
273
  cleaned = ' '.join(text.split())
254
274
  if self._current_row is not None:
255
- self._current_row.append(cleaned)
275
+ self._current_row.append((cleaned,) + self._cell_span)
256
276
  self._current_cell = None
257
277
  elif tag == 'tr' and self._current_row is not None:
258
278
  self._current_table.append(self._current_row)
@@ -263,14 +283,74 @@ class _TableExtractor(HTMLParser):
263
283
  if self._current_cell is not None:
264
284
  self._current_cell.append(data)
265
285
 
266
- def header_row_index(self) -> Optional[int]:
267
- """Index of the first row made of <th> cells, if there is one."""
268
- for index, is_header in enumerate(self._header_flags):
286
+ def header_row_index(self, table: int = 0) -> Optional[int]:
287
+ """Index of the first row made of <th> cells in one table, if any."""
288
+ for index, is_header in enumerate(self.header_flags[table]):
269
289
  if is_header:
270
290
  return index
271
291
  return None
272
292
 
273
293
 
294
+ def _span(attrs, name) -> int:
295
+ """A rowspan/colspan attribute as a sane positive int (default 1)."""
296
+ for key, value in attrs:
297
+ if key == name:
298
+ try:
299
+ return max(1, min(int(str(value).strip()), 1000))
300
+ except ValueError:
301
+ return 1
302
+ return 1
303
+
304
+
305
+ def _expand_spans(rows):
306
+ """
307
+ Lay ``(text, rowspan, colspan)`` cells out on a grid, as a browser does.
308
+
309
+ The cells a span covers are absent from the markup, so reading cells in
310
+ order slid every later value of those rows one column left. Covered
311
+ cells become '' and each span is returned as a block.
312
+ """
313
+ cells, spans = {}, []
314
+ for r, row in enumerate(rows):
315
+ c = 0
316
+ for text, rowspan, colspan in row:
317
+ while (r, c) in cells:
318
+ c += 1
319
+ rowspan = min(rowspan, len(rows) - r)
320
+ for dr in range(rowspan):
321
+ for dc in range(colspan):
322
+ cells[(r + dr, c + dc)] = text if dr == dc == 0 else ''
323
+ if rowspan > 1 or colspan > 1:
324
+ spans.append((r, c, r + rowspan - 1, c + colspan - 1))
325
+ c += colspan
326
+ out = []
327
+ for r in range(len(rows)):
328
+ width = max([c + 1 for (rr, c) in cells if rr == r] + [0])
329
+ out.append([cells.get((r, c), '') for c in range(width)])
330
+ return out, spans
331
+
332
+
333
+ _PAYLOAD_RE = re.compile(r'^var DATA = (\{.*\});$', re.MULTILINE)
334
+
335
+
336
+ def _flexablelib_payload(markup: str):
337
+ """The ``{headers, rows}`` JSON a ``to_html`` page embeds, or None."""
338
+ import json
339
+ match = _PAYLOAD_RE.search(markup)
340
+ if match is None:
341
+ return None
342
+ try:
343
+ payload = json.loads(match.group(1))
344
+ except ValueError:
345
+ return None
346
+ if not isinstance(payload, dict):
347
+ return None
348
+ headers, rows = payload.get('headers'), payload.get('rows')
349
+ if not isinstance(headers, list) or not isinstance(rows, list):
350
+ return None
351
+ return {'headers': headers, 'rows': rows}
352
+
353
+
274
354
  def from_html(table_class, source, index: int = 0, parse_numbers: bool = True,
275
355
  encoding: str = 'utf-8', **options):
276
356
  """
@@ -296,29 +376,49 @@ def from_html(table_class, source, index: int = 0, parse_numbers: bool = True,
296
376
 
297
377
  if not extractor.tables:
298
378
  raise ValueError('no <table> element found')
299
- if index >= len(extractor.tables):
379
+ if not -len(extractor.tables) <= index < len(extractor.tables):
300
380
  raise IndexError(
301
381
  f'table index {index} out of range; '
302
382
  f'the document has {len(extractor.tables)}'
303
383
  )
304
384
 
305
385
  rows = extractor.tables[index]
386
+ if len(extractor.tables) == 1 and len(rows) <= 1:
387
+ # ``Table.to_html`` renders its rows from an inline JSON payload, so
388
+ # the <tbody> is empty in the markup. Read that payload instead of
389
+ # returning a header-only table for our own export.
390
+ payload = _flexablelib_payload(markup)
391
+ if payload is not None:
392
+ return _new_table(table_class, payload['headers'],
393
+ payload['rows'], parse_numbers, **options)
306
394
  if not rows:
307
395
  return table_class(**options)
308
396
 
309
- header_index = extractor.header_row_index() if index == 0 else None
310
- if header_index is not None and header_index < len(rows):
311
- headers = rows[header_index]
312
- body = rows[:header_index] + rows[header_index + 1:]
313
- else:
314
- headers, body = rows[0], rows[1:]
397
+ header_index = extractor.header_row_index(index)
398
+ if header_index is None or header_index >= len(rows):
399
+ header_index = 0
400
+ headers = rows[header_index]
401
+ body = rows[:header_index] + rows[header_index + 1:]
315
402
 
316
403
  # Ragged rows are padded so every row reaches the column count.
317
404
  width = max([len(headers)] + [len(row) for row in body]) if body else len(headers)
318
405
  headers = list(headers) + [f'column {i + 1}' for i in range(len(headers), width)]
319
406
  body = [list(row) + [''] * (width - len(row)) for row in body]
320
407
 
321
- return _new_table(table_class, headers, body, parse_numbers, **options)
408
+ table = _new_table(table_class, headers, body, parse_numbers, **options)
409
+ # Restore rowspan/colspan blocks that lie wholly in the body as merges.
410
+ for first_row, first_col, last_row, last_col in extractor.spans[index]:
411
+ if first_row <= header_index <= last_row:
412
+ continue
413
+ shift = 1 if first_row > header_index else 0
414
+ first_row, last_row = first_row - shift, last_row - shift
415
+ if last_row >= table.row_count or last_col >= table.column_count:
416
+ continue
417
+ try:
418
+ table.merge_cells(first_row, first_col, last_row, last_col)
419
+ except ValueError:
420
+ continue
421
+ return table
322
422
 
323
423
 
324
424
  def from_pandas(table_class, dataframe, include_index: bool = False, **options):
@@ -442,6 +542,31 @@ def from_excel(table_class, path, sheet=None, has_header: bool = True,
442
542
  return table
443
543
 
444
544
 
545
+ def _split_pipe_row(body: str) -> List[str]:
546
+ """
547
+ Split a Markdown row on unescaped pipes and unescape ``\\|``.
548
+
549
+ ``to_markdown`` writes a literal pipe as ``\\|``; splitting on every
550
+ ``|`` cut such a cell in two and left a stray backslash behind, so the
551
+ table did not round-trip.
552
+ """
553
+ cells, current, i = [], [], 0
554
+ while i < len(body):
555
+ ch = body[i]
556
+ if ch == '\\' and i + 1 < len(body) and body[i + 1] == '|':
557
+ current.append('|')
558
+ i += 2
559
+ continue
560
+ if ch == '|':
561
+ cells.append(''.join(current))
562
+ current = []
563
+ else:
564
+ current.append(ch)
565
+ i += 1
566
+ cells.append(''.join(current))
567
+ return cells
568
+
569
+
445
570
  def from_markdown(table_class, source, parse_numbers: bool = True, **options):
446
571
  """
447
572
  Parse a GitHub-flavoured Markdown pipe table into a Table.
@@ -469,9 +594,9 @@ def from_markdown(table_class, source, parse_numbers: bool = True, **options):
469
594
  if '|' not in ln:
470
595
  continue
471
596
  body = ln[1:] if ln.startswith('|') else ln
472
- if body.endswith('|'):
597
+ if body.endswith('|') and not body.endswith('\\|'):
473
598
  body = body[:-1]
474
- rows_raw.append([c.strip() for c in body.split('|')])
599
+ rows_raw.append([c.strip() for c in _split_pipe_row(body)])
475
600
 
476
601
  if not rows_raw:
477
602
  return _new_table(table_class, [], [], parse_numbers, **options)